An examination of 3-5 word lexical bundles in L2 English academic writing: Identifying proficiency differences

Abstract

This study investigated the use of corpus-driven, fixed-form formulaic sequences (i.e., lexical bundles) in a collection of L2 English academic writing produced by L1 Japanese university students. Essays were sourced from a 1st year academic English writing program in Japan, where writers addressed the same writing prompt and produced their texts under similar conditions. A large collection of these essays was subsequently collected and graded by professional writing instructors using a modified version of the 5-point TOEFL rubric. Graded essays were divided into low (1-2) and high (4-5) proficiency groupings, with both corpora containing the same number of essays, as well as similar average word counts and total word counts. AntConc (Anthony, 2022) was then used to extract all 3–5-word lexical bundles meeting minimum frequency and range criteria. These raw results were sorted to remove overlap (e.g., shorter bundles contained within longer sequences) before being functionally categorized and analyzed for intra-group tendencies and inter-group differences. Major proficiency level differences included the finding that higher-level writers displayed more frequent use of lexical bundles as a general category (types and tokens). Furthermore, the analysis of individual lexical bundles was used to identify specific sequences that could be used in future teaching interventions to highlight which structures should be avoided/encouraged in order to promote perceptions of proficiency in L2 academic English writing. The methodological and pedagogical implications of this research will be discussed.

Keywords

L2 English writing; Lexical Bundles; Proficiency differences

References

Anthony, L. (2022). AntConc (Version 4.2.0) [Computer Software]. Tokyo, Japan: Waseda University. Available from https://www.laurenceanthony.net/software